We discuss the properties of subhalos in cluster-size halos, using a high-resolution statistical sample: the Rhapsody simulations introduced in Wu et al. (2012). We demonstrate that the criteria applied to select subhalos have significant impact on the inferred properties of the sample, including the scatter in the number of subhalos, the correlation between the subhalo number and formation time, and the shape of subhalos' spatial distribution and velocity structure. We find that the number of subhalos, when selected using the peak maximum circular velocity in their histories (a property expected to be closely related to the galaxy luminosity), is uncorrelated with the formation time of the main halo. This is in contrast to the previously reported correlation from studies where subhalos are selected by the current maximum circular velocity; we show that this difference is a result of the tidal stripping of the subhalos. We also find that the dominance of the main halo and the subhalo mass fraction are strongly correlated with halo concentration and formation history. These correlations are important to take into account when interpreting results from cluster samples selected with different criteria. Our sample also includes a fossil cluster, which is presented separately and placed in the context of the rest of the sample.